1 from __future__ import unicode_literals
6 from .common import InfoExtractor
14 class TechTVMITIE(InfoExtractor):
15 IE_NAME = 'techtv.mit.edu'
16 _VALID_URL = r'https?://techtv\.mit\.edu/(videos|embeds)/(?P<id>\d+)'
19 'url': 'http://techtv.mit.edu/videos/25418-mit-dna-learning-center-set',
20 'md5': '1f8cb3e170d41fd74add04d3c9330e5f',
24 'title': 'MIT DNA Learning Center Set',
25 'description': 'md5:82313335e8a8a3f243351ba55bc1b474',
29 def _real_extract(self, url):
30 mobj = re.match(self._VALID_URL, url)
31 video_id = mobj.group('id')
32 raw_page = self._download_webpage(
33 'http://techtv.mit.edu/videos/%s' % video_id, video_id)
34 clean_page = re.compile(r'<!--.*?-->', re.S).sub('', raw_page)
36 base_url = self._search_regex(
37 r'ipadUrl: \'(.+?cloudfront.net/)', raw_page, 'base url')
38 formats_json = self._search_regex(
39 r'bitrates: (\[.+?\])', raw_page, 'video formats')
40 formats_mit = json.loads(formats_json)
43 'format_id': f['label'],
44 'url': base_url + f['url'].partition(':')[2],
45 'ext': f['url'].partition(':')[0],
53 title = get_element_by_id('edit-title', clean_page)
54 description = clean_html(get_element_by_id('edit-description', clean_page))
55 thumbnail = self._search_regex(
56 r'playlist:.*?url: \'(.+?)\'',
57 raw_page, 'thumbnail', flags=re.DOTALL)
63 'description': description,
64 'thumbnail': thumbnail,
68 class MITIE(TechTVMITIE):
69 IE_NAME = 'video.mit.edu'
70 _VALID_URL = r'https?://video\.mit\.edu/watch/(?P<title>[^/]+)'
73 'url': 'http://video.mit.edu/watch/the-government-is-profiling-you-13222/',
75 'md5': '7db01d5ccc1895fc5010e9c9e13648da',
79 'title': 'The Government is Profiling You',
80 'description': 'md5:ad5795fe1e1623b73620dbfd47df9afd',
84 def _real_extract(self, url):
85 mobj = re.match(self._VALID_URL, url)
86 page_title = mobj.group('title')
87 webpage = self._download_webpage(url, page_title)
88 embed_url = self._search_regex(
89 r'<iframe .*?src="(.+?)"', webpage, 'embed url')
90 return self.url_result(embed_url, ie='TechTVMIT')